A rule-based robot grasp planner inspired by human behaviours

Robots are increasingly taking place in many everyday environments and industrial settings to carry out complex tasks that require high dexterity. However, obtaining high dexterity by robots is a challenge, especially when the object to manipulate is flexible and deformable. On the other hand humans grasp and manipulate effortlessly thanks to their experience gained over the years. Here we developed rule-based robot grasp planner designed based on studies on the human grasping behaviour. In particular, the grasp planner returns the appropriate grasp strategy in terms of grasp type, grasp location, and grasp dimension, according to the symbolic representation of the current scenario.